基于MapReduce的序列规则在推荐系统中的研究

Sequential patterns for recommendation system based on MapReduce

目前常用的个性化推荐系统模型通常是基于协同过滤或者是基于内容的,也有部分基于关联规则的。这些算法没有考虑事务间的顺序,然而在很多应用中这样的顺序很重要。文章提出了一种简易的基于序列模式的推荐模型,并且考虑到大规模数据的处理,结合了MapReduce编程模型。这种简易的推荐模型可以用来辅助通常的个性化推荐系统。

The most commonly used model of personalized recommendation system is usually based on collaborative filtering or content-based , but also partly based on association rules . The algorithm does not consider the sequence between transactions , but the sequence is important in many applications . This paper presents a simple model based on sequential patterns for recommenda-tion system , and taking into account the large-scale data processing , combined with the MapReduce programming model . This sim-ple model of recommendation can be used to assist the usual personalized recommendation system .